Research map / Continual learning / Continual fine-tuning of LLMs

NoTA: Normalized Tensor Adaptation for Parameter-Efficient Continual Learning

NeurIPS 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Low-rank and model-merging adaptation

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes NoTA, a normalized tensor adaptation method for parameter-efficient rehearsal-free class-incremental learning that addresses the effect of cumulative adapter update magnitude on forgetting.

Problem

Rehearsal-free class-incremental learning must learn new classes sequentially without stored data, where catastrophic forgetting is central; the effect of cumulative update magnitude in adapter-based methods is underexplored.

Approach

Adapter-based CIL stores and accumulates task-specific update matrices; NoTA studies the cumulative update magnitude and normalizes tensor adaptation to control it (abstract is truncated, so the exact mechanism is not stated).

Limitations

• Only the truncated abstract was available; full method and results could not be verified

Closest papers by meaning

• CL-LoRA: Continual Low-Rank Adaptation for Rehearsal-Free Class-Incremental Learning (CVPR 2025 · similarity 0.88)
• Learn more, but bother less: parameter efficient continual learning (NeurIPS 2024 · similarity 0.88)
• PaLoRA: Paced Low-Rank Adaptation for Continual Learning (NeurIPS 2026 · similarity 0.88)
• Adapter Merging with Centroid Prototype Mapping for Scalable Class-Incremental Learning (CVPR 2025 · similarity 0.87)
• STAR: Stability-Inducing Weight Perturbation for Continual Learning (ICLR 2025 · similarity 0.87)
• BiLoRA: Almost-Orthogonal Parameter Spaces for Continual Learning (CVPR 2025 · similarity 0.87)